On this page, you can find the abstracts of the students projects within the Active2Gether project that took place in the academic year of 2014-2015.
People have several important places in their lives which they often visit. Work, school, sports, their own home and the homes of family members are often among these places. Traveling between these locations can be done in several ways and people choose their travel methods based on different factors like distance and transportation availability. Sometimes these people have little choice, for instance if they have no car and it is too far to bike, but on other occasions they have several travel options from which they can choose.
The research in this project aims to develop the algorithms to be able to answer the following questions. What options do people choose to use? Do they choose for the easy and non-active travel option like public transport or the car, or do they choose for the more active travel option of the bike or foot? How can we determine how they travel and can their decision be influenced in the direction of the more active option?
Behavior of humans during a journey contains several patterns. Some of these patterns are related to their transport choice and their physical activity during the journey. Determining and studying these patterns for young adults resulted in advice which could be used during the development process of a lifestyle coaching application. The data, gathered with the Fitbit device and mobile GPS, is analysed with R, which gave insight in human commuting patterns. These patterns were the foundation for the following advice.
It is recommendable to increase the awareness of people about the diversity of transport options they could take. Their awareness is in most cases not optimal which therefore may hold people back to choose for more active transport. The coaching should be focused on short journeys done by car. These journeys do cover a distance which is currently already been covered by an average bike journey. The same occurs with most metro and bus journeys. People could be relatively effectively stimulated to take active travel instead due to the fact that the distance is not significantly different in these journeys. People take 35% less steps during days with relatively bad weather. These days have far more potential with respect to physical activity: for example, communicate with a weather API and help the user to predict if the weather is really an obstacle for using active transport.
The availability of public transport and machinery such as elevators and escalators discourages many people to perform daily physical activity like cycling, walking or using the stairs. Researchers of the Active2Gether project developed three automated algorithms in order to evaluate physical activity behavior in three domains: active transport, sport and stair climbing. The developed algorithms were evaluated in this research by looking at the correlation between the provided algorithm scores, the individual scores, i.e. the participants rating themselves, and two outsider scores, i.e. two outsiders rating the physical behavior of the participants. The following research question was answered: Are the scores derived by automated algorithms representative for the evaluation of physical behavior among young adults? There were two subquestions that helped to answer the research question: (i) Do the scores derived by the automated algorithms correspond to the self-report scores of the participants? (ii) Do the scores derived by automated algorithms correspond to scores given by outsiders? Sixteen participants between the age of 18 and 25 were involved in this project. The experiment was conducted in two weeks. Quantitative data was gathered in two ways: each respondent used a Fitbit device that tracked the number of stairs they had walked and filled in daily questionnaires about their transportation choice and time spent on sports.
Fysieke inactiviteit is een groot probleem onder de volwassenwereldbevolking omdat dit één van de meest voorkomende doodsoorzaken is. Met de grote opkomst van smartphones is het mogelijk om fysieke actviteit te bevorderen met behulp van bewegingsapps. Er is echter weinig bekend over de werking van deze bewegingsapps. In dit project wordt onderzocht welke technieken en functionaliteiten in deze bewegingsapps worden toegepast. Dit wordt gedaan aan de hand van een framework dat is opgesteld, hiermee kan er van elke app worden onderzocht hoeveel technieken en functionaliteiten deze bevatten. Er zijn ongeveer 200 apps geselecteerd aan de hand van criteria waaraan apps moeten voldoen willen ze behoren tot 'bewegingsapps', deze apps zijn gedowload en getest aan de hand van het framework. Hierdoor is er een lijst ontstaan met de apps met veel en weinig functionaliteiten en technieken. Dit is handig voor vervolgonderzoek, verder onderzoek kan zich richten op de effectiviteit van de bewegingsapps. Het is nu niet bekend of een app met veel features ook effectief is en ervoor zorgt dat er meer bewogen wordt.
Several aspects underlying lifestyle can spread through a social network. Social contagion by people in your social network influence your opinions and subsequently your behaviour. Your physical activity will be enhanced when connections with people with high levels of physical activities become stronger. Being able to predict how behaviour changes due to social influences could be useful in order to make changes in your environment to obtain desirable behaviour.
An experiment has been conducted with 20 participants of a social network, measuring mostly their physical activity levels and the strength of connections between people, in order to observe if there is a contagion effect due to the relationships in the network. The information obtained from this experiment has been used as input for a computational model that has been developed in earlier researches, which is able to make predictions on the change (increase or decrease) of physical activity during the experiment. The predicted directions of physical activity (increase or decrease) generated by the model are compared to the directions of the real data from the experiment over time. The model predicted the direction of change correctly in 80% up to 87% of the cases investigated.
This thesis was transformed into a scientific article, that was accepted to the international conference of Social Informatics 2015 under the title "Analysis and Evaluation of Social Contagion of Physical Activity in a Group of Young Adults".
The aim of this study was to research how the SWAN-framework can be used to develop a smartphone application that gives coaching advices based on real-time sensory input to encourage more physical activity for the users and if such an application increases the physical activity of the user. The SWAN-framework is especially developed for building context-aware applications. There are other applications that use sensors to encourage more physical activity, however, there are only 2 applications that give feedback to the user based on the user's context, and only 2 applications make use of adaptation techniques. The coaching advices in the application are 92 altered advices from the Active2Gether database. Only 22 advices are implemented. These advices are sent to the user based on input from the sensor implemented in the application. The application was tested by two test persons on two different smartphones. Each person received two advices, which were not followed up. A reason for this could be that deciding when to give advices is very difficult. Another reason could be that the advices are not personal and encouraging enough. The amount of test persons and the period of testing is too short to draw conclusions about if such an application increases the physical activity of the user. In future development, external sensors in the Active2Gether coaching application may result in more specific advice options. Another advantage of using external sensors is that it will detect more accurate environmental characteristics. More research on giving personal advices and giving encouraging advices may result in more effective advices. It would be interesting to study the result after extending the number of test persons and the test period.
During my bachelor project, I studied the existing Active2Gether behavioural model. My goal was to quantify the correlation coefficients between all the concepts that are connected. This experiment focused on the concepts 'outcome expectations', 'self-efficacy', 'self-regulation' and 'intentions', on which the participants were stimulated. The main question was: what are the correlation coefficients between the different concepts in the model? The hypotheses to this question were: (1) groups stimulated on a certain concept will outperform the other groups on that concept, (2) all groups will overall walk more, showing a rise in the amount of steps per day.
Via questionnaires at the beginning and the end of the experiment, the changes in the values of the concepts in the model were quantified. During the experiment, a Fitbit device tracked the number of steps, and shorter questionnaires (intermediate questionnaires) quantified the change in the concepts of interest over time. This way of testing did not yield workable results, as none of the groups stimulated on a certain concept outperforms the other groups on that concept. However, there is a rise in the amount of steps taken per day. For future work, bigger groups and a period of testing that does not include the end of a sporting season or a lot of holidays is needed. In addition, it is paramount that the stimulation messages will be more personalized.
Despite the fact, that the benefits of physical activity are well-known, only 60% of the Dutch adults meets the guidelines for healthy physical activity. Therefore, effective interventions are needed to increase levels of physical activity. For example, increasing levels of physical activity, can be achieved be promoting active transport. To do so, we need to know the underlying pathways, and to get insights into which determinants are associated with active transport. Therefore, the current study aimed to examine whether the Theory of Planned Behavior is applicable for active transport, i.e. active commuting. We analyzed data from the Amsterdam Growth and Health Longitudinal Study. A total of 324 adults completed a questionnaire assessing different aspects of active transport, including psychological factors. The results showed that intention is associated with active commuting. Furthermore, the Theory of Planned Behavior can be used to explain active commuting in Dutch adults.
In the current study, we were interested in the association between personality and physical activity. To do so, we used data from the Amsterdam Growth and Health Longitudinal Study. In this cohort, participants were first included at the age of 13 years, and attended follow-up measurements frequently. Consequently, we were able to examine whether physical activity and personality traits change over time, e.g. in from adolescence into young adulthood into adulthood. The analysis showed that some of personality traits are associated with levels of physical activity. Furthermore, the associations were different for men and women. These analyses were a first step in identifying longitudinal associations of personality and physical activity. The next step will be to explore whether clusters of personality traits are associated with physical activity.